email-xactions-to-todo/scrape.go

98 lines
2.3 KiB
Go

package main
import (
"errors"
"fmt"
"io/ioutil"
"log"
"net/mail"
"regexp"
"strings"
"github.com/google/uuid"
)
type scraper interface {
scrape(*mail.Message) ([]*Transaction, error)
}
type chaseScraper struct{}
type citiScraper struct{}
func Scrape(m *mail.Message) ([]*Transaction, error) {
scraper, err := buildScraper(m)
if err != nil {
return nil, err
}
return scraper.scrape(m)
}
func buildScraper(m *mail.Message) (scraper, error) {
subject := fmt.Sprint(m.Header["Subject"])
if !containsAny(subject, "transaction", "report", "Transaction") {
return nil, errors.New("cannot build scraper for subject " + subject)
}
log.Printf("%+v", m.Header)
from := fmt.Sprint(m.Header["From"])
if strings.Contains(from, "Chase") {
return newChaseScraper(), nil
}
if strings.Contains(from, "Citi") {
return newCitiScraper(), nil
}
return nil, errors.New("unknown sender: " + from)
}
func newChaseScraper() scraper {
return &chaseScraper{}
}
func newCitiScraper() scraper {
return &citiScraper{}
}
func containsAny(a string, b ...string) bool {
for i := range b {
if strings.Contains(a, b[i]) {
return true
}
}
return false
}
func (c *chaseScraper) scrape(m *mail.Message) ([]*Transaction, error) {
b, err := ioutil.ReadAll(m.Body)
if err != nil {
return nil, err
}
regexp := regexp.MustCompile(`A charge of \([^)]*\) (?P<amount>[\d\.]+) at (?P<account>.*) has been authorized`)
matches := regexp.FindSubmatch(b)
if len(matches) < 2 {
return nil, fmt.Errorf("no matches found: %+v: %s", matches, b)
}
results := make(map[string][]string)
for i, name := range regexp.SubexpNames() {
if i != 0 && name != "" {
results[name] = append(results[name], string(matches[i]))
}
}
if len(results) != 2 || len(results["amount"]) != len(results["account"]) {
return nil, fmt.Errorf("unexpected matches found looking for transactions: %+v", results)
}
transactions := make([]*Transaction, len(results["amount"]))
for i := range results["amount"] {
transactions[i] = &Transaction{
Amount: results["amount"][i],
Account: results["account"][i],
Bank: Chase,
Date: fmt.Sprint(m.Header["Date"]),
ID: uuid.New().String(), // TODO random based on date, amount, account, and bank so find dupes
}
}
return transactions, nil
}
func (c *citiScraper) scrape(m *mail.Message) ([]*Transaction, error) {
panic("not impl")
}